1 /* cat -- concatenate files and print on the standard output.
2 Copyright (C) 1988, 1990, 1991 Free Software Foundation, Inc.
4 This program is free software; you can redistribute it and/or modify
5 it under the terms of the GNU General Public License as published by
6 the Free Software Foundation; either version 2, or (at your option)
9 This program is distributed in the hope that it will be useful,
10 but WITHOUT ANY WARRANTY; without even the implied warranty of
11 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 GNU General Public License for more details.
14 You should have received a copy of the GNU General Public License
15 along with this program; if not, write to the Free Software
16 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
18 /* Differences from the Unix cat:
19 * Always unbuffered, -u is ignored.
20 * 100 times faster with -v -u.
21 * 20 times faster with -v.
23 By tege@sics.se, Torbjorn Granlund, advised by rms, Richard Stallman. */
27 #include <sys/types.h>
29 #include <sys/ioctl.h>
34 #define max(h,i) ((h) > (i) ? (h) : (i))
41 static void next_line_num ();
42 static void simple_cat ();
44 /* Name under which this program was invoked. */
47 /* Name of input file. May be "-". */
50 /* Descriptor on which input file is open. */
51 static int input_desc;
53 /* Descriptor on which output file is open. Always is 1. */
54 static int output_desc;
56 /* Buffer for line numbers. */
57 static char line_buf[13] =
58 {' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', '0', '\t', '\0'};
60 /* Position in `line_buf' where printing starts. This will not change
61 unless the number of lines are more than 999999. */
62 static char *line_num_print = line_buf + 5;
64 /* Position of the first digit in `line_buf'. */
65 static char *line_num_start = line_buf + 10;
67 /* Position of the last digit in `line_buf'. */
68 static char *line_num_end = line_buf + 10;
70 /* Preserves the `cat' function's local `newlines' between invocations. */
71 static int newlines2 = 0;
73 /* Count of non-fatal error conditions. */
74 static int exit_stat = 0;
80 Usage: %s [-benstuvAET] [--number] [--number-nonblank] [--squeeze-blank]\n\
81 [--show-nonprinting] [--show-ends] [--show-tabs] [--show-all]\n\
82 [--help] [--version] [file...]\n",
94 /* Optimal size of i/o operations of output. */
97 /* Optimal size of i/o operations of input. */
100 /* Pointer to the input buffer. */
101 unsigned char *inbuf;
103 /* Pointer to the output buffer. */
104 unsigned char *outbuf;
108 /* Index in argv to processed argument. */
111 /* Device number of the output (file or whatever). */
114 /* I-node number of the output. */
117 /* Nonzero if the output file should not be the same as any input file. */
118 int check_redirection = 1;
120 /* Nonzero if we have ever read standard input. */
121 int have_read_stdin = 0;
123 struct stat stat_buf;
125 /* Variables that are set according to the specified options. */
127 int numbers_at_empty_lines = 1;
128 int squeeze_empty_lines = 0;
129 int mark_line_ends = 0;
133 /* If non-zero, call cat, otherwise call simple_cat to do the actual work. */
136 /* If non-zero, display usage information and exit. */
137 static int flag_help;
139 /* If non-zero, print the version on standard error. */
140 static int flag_version;
142 static struct option const long_options[] =
144 {"number-nonblank", no_argument, NULL, 'b'},
145 {"number", no_argument, NULL, 'n'},
146 {"squeeze-blank", no_argument, NULL, 's'},
147 {"show-nonprinting", no_argument, NULL, 'v'},
148 {"show-ends", no_argument, NULL, 'E'},
149 {"show-tabs", no_argument, NULL, 'T'},
150 {"show-all", no_argument, NULL, 'A'},
151 {"help", no_argument, &flag_help, 1},
152 {"version", no_argument, &flag_version, 1},
156 program_name = argv[0];
158 /* Parse command line options. */
160 while ((c = getopt_long (argc, argv, "benstuvAET", long_options, (int *) 0))
171 numbers_at_empty_lines = 0;
187 squeeze_empty_lines = 1;
197 /* We provide the -u feature unconditionally. */
229 fprintf (stderr, "%s\n", version_string);
238 /* Get device, i-node number, and optimal blocksize of output. */
240 if (fstat (output_desc, &stat_buf) < 0)
241 error (1, errno, "standard output");
243 outsize = ST_BLKSIZE (stat_buf);
244 /* Input file can be output file for non-regular files.
245 fstat on pipes returns S_IFSOCK on some systems, S_IFIFO
246 on others, so the checking should not be done for those types,
247 and to allow things like cat < /dev/tty > /dev/tty, checking
248 is not done for device files either. */
250 if (S_ISREG (stat_buf.st_mode))
252 out_dev = stat_buf.st_dev;
253 out_ino = stat_buf.st_ino;
256 check_redirection = 0;
258 /* Check if any of the input files are the same as the output file. */
268 infile = argv[argind];
270 if (infile[0] == '-' && infile[1] == 0)
277 input_desc = open (infile, O_RDONLY);
280 error (0, errno, "%s", infile);
286 if (fstat (input_desc, &stat_buf) < 0)
288 error (0, errno, "%s", infile);
292 insize = ST_BLKSIZE (stat_buf);
294 /* Compare the device and i-node numbers of this input file with
295 the corresponding values of the (output file associated with)
296 stdout, and skip this input file if they coincide. Input
297 files cannot be redirected to themselves. */
299 if (check_redirection
300 && stat_buf.st_dev == out_dev && stat_buf.st_ino == out_ino)
302 error (0, 0, "%s: input file is output file", infile);
307 /* Select which version of `cat' to use. If any options (more than -u,
308 --version, or --help) were specified, use `cat', otherwise use
313 insize = max (insize, outsize);
314 inbuf = (unsigned char *) xmalloc (insize);
316 simple_cat (inbuf, insize);
320 inbuf = (unsigned char *) xmalloc (insize + 1);
322 /* Why are (OUTSIZE - 1 + INSIZE * 4 + 13) bytes allocated for
325 A test whether output needs to be written is done when the input
326 buffer empties or when a newline appears in the input. After
327 output is written, at most (OUTSIZE - 1) bytes will remain in the
328 buffer. Now INSIZE bytes of input is read. Each input character
329 may grow by a factor of 4 (by the prepending of M-^). If all
330 characters do, and no newlines appear in this block of input, we
331 will have at most (OUTSIZE - 1 + INSIZE) bytes in the buffer. If
332 the last character in the preceding block of input was a
333 newline, a line number may be written (according to the given
334 options) as the first thing in the output buffer. (Done after the
335 new input is read, but before processing of the input begins.) A
336 line number requires seldom more than 13 positions. */
338 outbuf = (unsigned char *) xmalloc (outsize - 1 + insize * 4 + 13);
340 cat (inbuf, insize, outbuf, outsize, quote,
341 output_tabs, numbers, numbers_at_empty_lines, mark_line_ends,
342 squeeze_empty_lines);
350 if (strcmp (infile, "-") && close (input_desc) < 0)
352 error (0, errno, "%s", infile);
356 while (++argind < argc);
358 if (have_read_stdin && close (0) < 0)
359 error (1, errno, "-");
361 error (1, errno, "write error");
366 /* Plain cat. Copies the file behind `input_desc' to the file behind
370 simple_cat (buf, bufsize)
371 /* Pointer to the buffer, used by reads and writes. */
374 /* Number of characters preferably read or written by each read and write
378 /* Actual number of characters read, and therefore written. */
381 /* Loop until the end of the file. */
385 /* Read a block of input. */
387 n_read = read (input_desc, buf, bufsize);
390 error (0, errno, "%s", infile);
395 /* End of this file? */
400 /* Write this block out. */
402 if (write (output_desc, buf, n_read) != n_read)
403 error (1, errno, "write error");
407 /* Cat the file behind INPUT_DESC to the file behind OUTPUT_DESC.
408 Called if any option more than -u was specified.
410 A newline character is always put at the end of the buffer, to make
411 an explicit test for buffer end unnecessary. */
414 cat (inbuf, insize, outbuf, outsize, quote,
415 output_tabs, numbers, numbers_at_empty_lines,
416 mark_line_ends, squeeze_empty_lines)
418 /* Pointer to the beginning of the input buffer. */
419 unsigned char *inbuf;
421 /* Number of characters read in each read call. */
424 /* Pointer to the beginning of the output buffer. */
425 unsigned char *outbuf;
427 /* Number of characters written by each write call. */
430 /* Variables that have values according to the specified options. */
434 int numbers_at_empty_lines;
436 int squeeze_empty_lines;
438 /* Last character read from the input buffer. */
441 /* Pointer to the next character in the input buffer. */
444 /* Pointer to the first non-valid byte in the input buffer, i.e. the
445 current end of the buffer. */
448 /* Pointer to the position where the next character shall be written. */
449 unsigned char *bpout;
451 /* Number of characters read by the last read call. */
454 /* Determines how many consecutive newlines there have been in the
455 input. 0 newlines makes NEWLINES -1, 1 newline makes NEWLINES 1,
456 etc. Initially 0 to indicate that we are at the beginning of a
457 new line. The "state" of the procedure is determined by
459 int newlines = newlines2;
462 /* If nonzero, use the FIONREAD ioctl, as an optimization.
463 (On Ultrix, it is not supported on NFS filesystems.) */
464 int use_fionread = 1;
467 /* The inbuf pointers are initialized so that BPIN > EOB, and thereby input
468 is read immediately. */
479 /* Write if there are at least OUTSIZE bytes in OUTBUF. */
481 if (bpout - outbuf >= outsize)
483 unsigned char *wp = outbuf;
486 if (write (output_desc, wp, outsize) != outsize)
487 error (1, errno, "write error");
490 while (bpout - wp >= outsize);
492 /* Move the remaining bytes to the beginning of the
495 bcopy (wp, outbuf, bpout - wp);
496 bpout = outbuf + (bpout - wp);
499 /* Is INBUF empty? */
506 /* Is there any input to read immediately?
507 If not, we are about to wait,
508 so write all buffered output before waiting. */
511 && ioctl (input_desc, FIONREAD, &n_to_read) < 0)
513 /* Ultrix returns EOPNOTSUPP on NFS;
514 HP-UX returns ENOTTY on pipes. */
515 if (errno == EOPNOTSUPP || errno == ENOTTY)
519 error (0, errno, "cannot do ioctl on `%s'", infile);
521 newlines2 = newlines;
528 int n_write = bpout - outbuf;
530 if (write (output_desc, outbuf, n_write) != n_write)
531 error (1, errno, "write error");
535 /* Read more input into INBUF. */
537 n_read = read (input_desc, inbuf, insize);
540 error (0, errno, "%s", infile);
542 newlines2 = newlines;
547 newlines2 = newlines;
551 /* Update the pointers and insert a sentinel at the buffer
560 /* It was a real (not a sentinel) newline. */
562 /* Was the last line empty?
563 (i.e. have two or more consecutive newlines been read?) */
567 /* Are multiple adjacent empty lines to be substituted by
568 single ditto (-s), and this was the second empty line? */
570 if (squeeze_empty_lines && newlines >= 2)
576 /* Are line numbers to be written at empty lines (-n)? */
578 if (numbers && numbers_at_empty_lines)
581 bpout = (unsigned char *) stpcpy (bpout, line_num_print);
585 /* Output a currency symbol if requested (-e). */
590 /* Output the newline. */
598 /* Are we at the beginning of a line, and line numbers are requested? */
600 if (newlines >= 0 && numbers)
603 bpout = (unsigned char *) stpcpy (bpout, line_num_print);
606 /* Here CH cannot contain a newline character. */
608 /* The loops below continue until a newline character is found,
609 which means that the buffer is empty or that a proper newline
612 /* If quoting, i.e. at least one of -v, -e, or -t specified,
613 scan for chars that need conversion. */
636 *bpout++ = ch - 128 + 64;
639 else if (ch == '\t' && output_tabs)
653 /* Not quoting, neither of -v, -e, or -t specified. */
656 if (ch == '\t' && !output_tabs)
672 /* Compute the next line number. */
677 char *endp = line_num_end;
684 while (endp >= line_num_start);
685 *--line_num_start = '1';
686 if (line_num_start < line_num_print)